Reality Check: Are We in an AI Bubble?

Sam Altman of OpenAI recently got a lot of press when he said in a conversation that it feels like we might be in an AI bubble. For folks in the construction industry, this means something different to someone just watching the news, it makes us wonder if the data center boom might come to an abrupt end. 

First, let’s look at the startup world, where Sam spent most of his career prior to OpenAI. I don’t think anyone would dispute that yes, the startup and VC world is in a bubble when it comes to AI. Whether it bursts or just deflates we will see, but when everyone’s piling into one type of business, we seem to see bubbles happen. 

Second, and more complicated, is whether the broader economy is in an AI bubble. The fact that the stock market is reaching new highs led by the handful of companies who are spending billions on data centers is concerning. These valuations, and this spending, make sense only if AI truly transforms productivity across the economy, and the (very) early evidence is mixed - a flawed but famous MIT study showed that 95% of AI agent deployments are showing zero ROI, but at the same time around a billion people are using chatbots at least weekly. So, there is something there. Another point in favor of it not being a dangerous bubble is most of the AI infrastructure spending is from company profits and cashflow, vs.  direct investor funding, which might mean some big companies book losses but need not mean the economy collapses if the AI promises turn out to be overestimated or take too long to show up. 

The Reality Check - we don’t know if this is a dangerous, economy-wide bubble. We do know that for at least the next 18 months, the cash driving AI investment is likely to continue, and the drive to make AI useful is relentless. Let’s see what it looks like in September 2026.